J i on Electr c o u a rn l o f P r ob abil ity Vol. 6 (2001) Paper no. 13, pages 1–27 Journal URL http://www.math.washington.edu/~ejpecp/ Paper URL http://www.math.washington.edu/~ejpecp/EjpVol6/paper13.abs.html BOUNDARY CONDITIONS FOR ONE-DIMENSIONAL BIHARMONIC PSEUDO PROCESS Kunio Nishioka Department of Mathematics, Tokyo Metropolitan University Minami-Ohsawa, Hachioji-shi, Tokyo 192-0397, Japan kunio@comp.metro-u.ac.jp Abstract We study boundary conditions for a stochastic pseudo processes corresponding to the biharmonic operator. The biharmonic pseudo process (BP P for short). is composed, in a sense, of two different particles, a monopole and a dipole. We show how an initial-boundary problems for a 4-th order parabolic differential equation can be represented by BP P with various boundary conditions for the two particles: killing, reflection and stopping. Keywords Boundary conditions for biharmonic pseudo process, killing, reflection, stopping AMS Subject Classification Primary. 60J50; Secondary. 35K35, 60G18, 60G20, 60J30. Submitted to EJP on September 6, 2000. Final version accepted on May 21, 2001. 1 Introduction We will call −42 ≡ −∂x4 a biharmonic operator. It plays an important role in the theory of elasticity and fluid dynamics. For instance the parabolic differential equation ∂t u(t, x) = −∂x4 u(t, x), t > 0, x ∈ R1 , (1.1) is closely related to the Kuramoto-Sivashinsky equation and the Cahne-Hilliard equation (see [17, Chapter III §4]). It is easy to see that the fundamental solution p(t, x) to (1.1) is given by Z ∞ 1 p(t, x) = dξ exp{−iξx − ξ 4 t}, t > 0, x ∈ R 1 . (1.2) 2π −∞ This p(t, x) takes negative values (see Hochberg [8]), so no stochastic process corresponds to (1.1) in the usual sense. Several attempts have been made to relate the biharmonic operator to random processes. Krylov [9] considered a stochastic pseudo process whose “transition probability density” was p(t, x) as in (1.2), despite its taking negative values. We will call this pseudo process a biharmonic pseudo process or BP P for short (it is occasionally called the Krylov motion). Following Krylov’s idea, Hochberg [8] started a systematic study of BP P . His article included a definition of a stochastic integral with respect to BP P . Nishioka [14] calculated the joint distribution of the first hitting time and place for BP P hitting the boundary of a half-line. This was extended by Nakajima and Sato [10] to the space-time case. Some new developments in this direction can be found in a paper by Beghin, Hochberg and Orsingher [1]. Other attempts included a paper by Funaki [7] who introduced the concept of iterated Brownian motion. After his pioneering work, Burdzy [2] began to study path properties of iterated Brownian motion; a large number of related papers followed. Burdzy and Ma̧drecki [3, 4] modified Funaki’s model in a different way; they defined a stochastic integral with respect to the their process. This paper is devoted to the process BP P on a half-line. We will study various boundary conditions for the probabilistic model and relate them to their analytic counterparts. We will motivate our results by first reviewing the classical case of the standard Brownian motion on (0, ∞). Consider the following initial-boundary value problem for the heat equation: ∂t u(t, x) = (1/2) ∂x2 u(t, x), t > 0, x > 0; ∂x` u(t, 0) = 0, u(0, x) = f (x), x > 0; t > 0, (1.3) where ` may take values 0, 1 or 2. Solutions to this initial-boundary value problem may be represented using Brownian motion on (0, ∞) which is killed, reflected, or stopped at the boundary point 0, depending on the value of `. The relationship can be summarized as follows: ` 0 1 2 Table 1.1. Brownian boundary condition killing reflection stopping 2 We note that there is no analytic difference between Brownian transition probabilities on the open interval (0, ∞) in the cases when the Brownian particle is killed or stopped at the boundary but we make the distinction in Table 1.1 to emphasize the analogy with some results on BP P . In this article, we will establish a relationship analogous to Table 1.1, between BP P and several initial-boundary value problems for (1.1). We start with a few heuristic ideas. A big difference between a Brownian motion and BP P , proved in [13], is that the Brownian motion is a model for the motion of a single particle but when the BP P leaves the interval (0, ∞), it appears to consist of two types of particles. We will call these particles a monopole and a dipole. The names are justified by a result on the conservation of charges given in the last section of this article. The particles behave independently at the boundary. Therefore we need two different boundary conditions in order to solve the initialboundary value problem for (1.1). In a sense, each boundary condition controls the behavior of a monopole or a dipole, although the exact relationship, summarized in Table 1.2 below, is more complicated than that. We proceed with a more formal presentation of the main results although the fully rigorous statements are postponed until later in the paper. Consider the following initial-boundary value problems for (1.1), ∂t v(t, x) = −∂x4 v(t, x), t > 0, x > 0; ∂x` v(t, 0) = 0 and ∂xm v(t, 0) v(0, x) = f (x), x > 0; = 0, t > 0, (1.4) where f is a given bounded smooth function on [0, ∞) and ` and m can take the values ` = 0, . . . , 4 and m = 0, . . . , 5. The following two tables summarize our main results. `=0 `=1 `=2 `=3 `=4 m=0 −− −− −− −− −− m=1 00 −− −− −− −− Table 1.2. m=2 m=3 0r H r0 H −− H −− −− −− −− m=4 × s0 sr H −− m=5 0s × H H ss For some values of ` and m, the initial boundary value problem has a probabilistic representation using BP P with killing, reflection or stopping for the monopole and dipole. The correspondence between the various values of ` and m and boundary conditions for BP P is coded in Tables 1.2 and 1.3 in the self-evident manner. killing for monopole reflection for monopole stopping for monopole Table 1.3. killing for dipole reflection for dipole 00 0r (see §4) (see §6) r0 −− (see §5.1) (see §5.3) s0 sr (see §7.2) (see §7.3) 3 stopping for dipole 0s (see §8.1) −− (see §8.2) ss (see §7.1) In Table 1.2, the symbol “×” means that (1.4) is not well-posed with such analytic boundary conditions (see Example 7.9). The symbol “H” means such analytic boundary conditions are not related to BP P with killing, reflection or stopping at the boundary. However, these analytic conditions can be related to BP P . This requires a new idea of a “higher order reflecting boundary” which will be discussed in a forthcoming article. The diagonal entries in Table 1.2 and those below the diagonal are irrelevant for obvious reasons and so they are marked with “−−”. The same symbol in Table 1.3 means that we have not found a relationship between this set of analytic conditions and BP P . In §2, we review results from [13, 14] on the joint distribution of the first hitting time and place for BP P and the concepts of a monopole and dipole. In §3 through §8, we construct and study BP P ’s with various boundary conditions. The construction of the BP P in the case when we have killing or stopping of the monopole at the boundary is similar in a sense to the Brownian motion case. Only slight modifications are needed when the same boundary conditions are prescribed to the dipole. When we set the reflecting boundary for a monopole or dipole, we cannot apply a direct analogy with the reflecting Brownian motion (see §5.2). We will construct a BP P with the reflection as the limit of a suitable sequence of approximations. In §9, we verify the law of the conservation of charges, and conclude that the total amount of charges is conservative if and only if there is no killing for a monopole. We will use some basic results on solutions of 4-th order PDE’s, Laplace and Fourier-Laplace transforms through the paper. These can be found in [5, 6, 16] for instance. Acknowlegement The author would like to thank the referee made many important and precise suggestions for improvements of the paper. 2 Preliminaries We start with a review of several function spaces used in the article. Let X denote either the one dimensional Euclidean space R 1 or the half-line [0, ∞). Bb (X) will stand for the space of all bounded measurable functions defined on X. C(X) will be the space of all continuous functions on X while Cb (X) will be the space of all bounded functions in C(X). C1 (X) will denote the space of all continuously differentiable functions on X and C1b (X) will be the space of all bounded functions in C1 (X). D[0, ∞) will denote the space of all right continuous functions defined on [0, ∞) which have left hand limits. S will be the space of all Schwartz class functions defined on R1 . Dirac’s delta function will be denoted by δ(x) and δa (db) will be the delta measure with the unit mass at a point {a}, so we may use the convention δa (db) = δ(a − b) db. For a fixed positive t, the function p(t, x) of (1.2) is an even function of x and it belongs to S. 4 For positive t and s, the following formulae hold Z ∞ x dx p(t, x) = 1, p(t, x) = t−1/4 p 1, 1/4 , t −∞ Z ∞ and dy p(t, y − x) p(s, y) = p(t + s, x). (2.1) −∞ Note that p(t, x) can take negative values. Hochberg [8] proved that p(1, |x|) = a|x|−1/3 exp{−b|x|4/3 } cos c|x|4/3 + a lower order term for large |x|, where a, b, and c are positive constants. It follows that Z ∞ Z ∞ |p(t, x)| dx = |p(1, x)| dx = a constant > 1 for any t > 0. −∞ (2.2) −∞ e x on cylinder sets in Using this p(t, x), Krylov [9] defined a finitely additive signed measure P [0,∞) [0,∞) R . A cylinder set in R , say Γ, is a set such that Γ = {ω ∈ R [0,∞) : ω(t1 ) ∈ B1 , · · · , ω(tn ) ∈ Bn } where 0 ≤ t1 < · · · < tn and Bk ’s are Borel sets in R 1 . We put Z Z e x [Γ] ≡ P dy1 · · · dyn p(t1 , y1 − x)× B1 (2.3) (2.4) Bn × p(t2 − t1 , y1 − y2 ) · · · p(tn − tn−1 , yn − yn−1 ). e x is a σ–additive finite measure on R n , but it is not a If we fix 0 ≤ t1 < · · · < tn , then this P σ–additive measure on the smallest σ–field which includes all cylinder sets in R [0,∞) . We say that a function f defined on R[0,∞) is tame, if f (ω) = g(ω(t1 ), · · · , ω(tn )), ω ∈ R [0,∞) , (2.5) for a Borel function g on R n and 0 ≤ t1 < · · · < tn . For each tame function f , we define its expectation in the usual way—if f is as in (2.5), then we set Z e e x [dω(t1 ) × · · · × dω(tn )], Ex [f (ω)] = f (ω) P (2.6) if the right hand side exists. We extend the expectation to other functions as follows. Let n and N be natural numbers. For each ω ∈ R[0,∞) , we set if k/2n ≤ t < (k + 1)/2n and t < N ω(k/2n ) ωnN (t) ≡ ω(N ) if t ≥ N . Definition 2.1. We say that a function F defined on following conditions, 5 R[0,∞) is admissible if F satisfies the (i) for each n and N , F (ωnN ) is tame, (ii) for each ω ∈ R[0,∞) , limn→∞ limN →∞ F (ωnN ) = F (ω), P k k−1 e e (iii) there exists limn→∞ limN →∞ N k=1 |Ex [F (ωn )] − Ex [F (ωn )]|. We define the expectation of an admissible function F (ω) by e x [F (ω N )]. Ex [F (ω)] ≡ lim lim E n n→∞ N →∞ The expectation is unique if it exists, due to (iii) of Definition 2.1 (see [14]). Let A be a subset in R [0,∞) . When the characteristic function IA (ω) is admissible, we let Px [A] ≡ Ex [IA (ω)]. (2.7) For a positive number α, Hα [0, ∞) will denote the space of all functions defined on [0, ∞) which satisfy Hölder’s condition of order α. According to Krylov [9], Px –total variation of Hα [0, ∞) c = 0 if α < 14 . This implies that the total mass of |Px | is concentrated on C[0, ∞). However for a technical reason related to Definition 2.1 (we need to deal with ωnN ), we will take a larger space D[0, ∞) as the path space of the pseudo process corresponding to Px . From now on, we will identify Px and Ex with their restrictions to D[0, ∞). Definition 2.2. A biharmonic pseudo process, or BP P in short, is the family of finitely additive signed measures {Px : x ∈ R1 }, defined in (2.4) and (2.7). The domain of Px is a finitely additive field in D[0, ∞) which includes all cylinder sets. 3 The first hitting time distribution, monopole and dipole We will review results from [13, 14] in this section.. 3.1 The distribution of the first hitting time and place Given ω ∈ D[0, ∞), the random time τ0 (ω) ≡ inf{t > 0 : ω(t) < 0}, is called the first hitting time of the interval (−∞, 0). It can be proved that the function exp{−λτ0 (ω) + iβω(τ0 )} (3.1) is admissible, and we can calculate its expectation. For each λ > 0 and β ∈ R 1 , Ex [exp{−λτ0 (ω) + iβω(τ0 )}] 1 = √ θ1 exp{λ1/4 θ2 x} + θ1 exp{λ1/4 θ2 x} 2 iβ + √ −i exp{λ1/4 θ2 x} + i exp{λ1/4 θ2 x} , 1/4 2λ 6 (3.2) where θ1 ≡ exp{πi/4} and θ2 ≡ exp{3πi/4}. In the classical probability setting, one can derive the joint distribution of the first hitting time and place Px [τ0 (ω) ∈ dt, ω(τ0 ) ∈ da] (3.3) from (3.2), using Bochner’s theorem. However the theorem cannot be applied to BP P , since (3.2) is not positive definite in β, and so we have to extend the notion of the “joint distribution” itself. Given functions φ and ϕ in S, it can be proved that the function exp{−λτ0 (ω)} φ(τ0 (ω)) ϕ(ω(τ0 )) is admissible and its expectation defines a continuous bilinear functional on S ×S. This fact is directly implied by Schwartz’s result on Fourier-Laplace transforms of his temperate distributions. The following lemma is a slight modification of his result. Lemma 3.1 ([15]). Let c be a positive number and U (λ, β) be a complex-valued function of a real variable β and a complex variable λ with <λ > c. Assume that for each real β, U is a holomorphic function in λ with <λ > c and it satisfies k |U (λ, β)| ≤ C 1 + |λ| + |β| for <λ > c and β ∈ R 1 , (3.4) where C and k are non-negative constants. Then U is the Fourier-Laplace transform of a Schwartz temperate distribution whose support lies in [0, ∞) × R 1 . We will always take the principal value for λ1/4 . If <λ > c for any positive number c, then −π/8 < arg λ1/4 < π/8. This implies that (3.2) satisfies (3.4). Hence there exists a Schwartz temperate distribution q(t, a; x) such that Ex [exp{−λτ0 (ω)} φ(τ0 (ω)) ϕ(ω(τ0 ))] Z ∞ Z = dt da exp{−λt} q(t, a; x) φ(t) ϕ(a), λ > c > 0, (3.5) 0 holds for all φ and ϕ in S. In the classical probability theory, distributions of real-valued random variables may be thought of as non-negative continuous linear functionals on the function space Cb (R 1 ). We define distributions of functions of BP P to be continuous linear functionals on a function space which includes S. Definition 3.2. We call Schwartz’s temperate distribution q(t, a; x) in (3.5) the density of the distribution (3.3), and let Px [τ0 (ω) ∈ dt, ω(τ0 ) ∈ da] ≡ q(t, a; x) dt da, which is a continuous bilinear functional on S × S for each x ≥ 0. Proposition 3.3 ([14]). Let x ≥ 0. Then in the distribution sense, Px [τ0 (ω) ∈ dt, ω(τ0 ) ∈ da] = K(t, x) δ(a) − J(t, x) δ0 (a) dt da 7 (3.6) where δ0 (a) is the first derivative of Dirac’s delta function δ(a), Z 2 ∞ K(t, x) ≡ dy exp{−y 4 t} y 3 sin yx − cos yx + exp{−yx} , π 0 Z 2 ∞ J(t, x) ≡ dy exp{−y 4 t} y 2 sin yx − cos yx + exp{−yx} . π 0 (3.7) Moreover, the support of the distribution in (3.6) is [0, ∞) × {0}. Remark 3.4 ([14]). (i) For each t > 0, K(t, x) and J(t, x) are C∞ b [0, ∞) functions in x. For each x > 0, they are also C∞ [0, ∞) functions in t. b (ii) For some positive constant C independent of t and x, we have !n C x2 t1/4 |K(t, x)| ≤ for n = 0, · · · , 6; x t3/2 !n C x2 t1/4 |J(t, x)| ≤ for n = 0, · · · , 5. x t5/4 (iii) For x > 0, Z Z ∞ K(t, x) dt = 1 ∞ and 0 J(t, x) dt = x. 0 (iv) Let f be a bounded continuous function on [0, ∞). For t > 0, Z lim x→0 0 Z t ds K(s, x) f (s) = f (0) and lim x→0 0 t ds J(s, x) f (s) = 0. The explicit formula (3.6) makes it possible to extend (3.3) to a continuous bilinear functional on a larger space than S × S. Corollary 3.5 ([14]). The distribution (3.3) can be extended to a continuous bilinear functional on Bb [0, ∞) × C1 (R 1 ). The strong Markov property holds for BP P at time τ0 (ω). Proposition 3.6 (Strong Markov Property [14]). Let y < 0 < x. The following holds in the sense of continuous linear functionals on Bb [0, ∞), Z Px [ω(t) ∈ dy] = 3.2 t Z s=0 ∞ a=−∞ Px [τ0 (ω) ∈ ds, ω(τ0 ) ∈ da] Pa [ω(t − s) ∈ dy]. (3.8) Monopole and dipole In physics a particle is called a dipole when it carries two charges of equal magnitude but opposite signs. A small magnet is a typical example of a dipole. Heuristically, one may represent a dipole by 1 1 − δ(a + ) + δ(a − ). (3.9) 2 2 8 As → 0, (3.9) converges to −δ0 (a) in the distribution sense. Therefore we will call −δ0 (a) a dipole. For the Brownian motion B(t), it is well known that x Px [τ0 ∈ dt, B(τ0 ) ∈ da] = √ exp{−x2 /2t} δ(a) dt da; x > 0. 2πt3 Comparing this with (3.6), we see that in a sense, BP P behaves as a mixture of two particles of different types when it hits an interval. Informally speaking, a particle of the first type is represented by δ(a) and carries a charge of a single sign just as a Brownian particle does. The second particle is represented by −δ0 (a) and corresponds to a dipole. The following informal definition will enable us to present our results using intuitive notation. Definition 3.7. We will call a particle represented by δ(a) a monopole and a particle represented by −δ0 (a) a dipole. We define distributions of the first hitting time and place for the monopole and dipole as follows. Px [τ0 (ω) ∈ dt, ω(τ0 ) is a monopole and in da] ≡ K(t, x) δ(a) dt da, Px [τ0 (ω) ∈ dt, ω(τ0 ) is a dipole and in da] ≡ J(t, x) (−δ0 (a)) dt da, where the first distribution is a continuous linear functional on Bb [0, ∞) × C(R1 ) and the latter is a functional on Bb [0, ∞) × C1 (R 1 ). Corollary 3.8. In the sense of continuous linear functionals on Bb [0, ∞) × C1 (R 1 ), Pb [τ0 (ω) ∈ dt, ω(τ0 ) ∈ da] = Px [τ0 (ω) ∈ dt, ω(τ0 ) is a monopole and in da] (3.10) + Px [τ0 (ω) ∈ dt, ω(τ0 ) is a dipole and in da]. 3.3 The initial particle When BP P starts with an initial distribution µ(dy), its distribution at time t is given by Z µ(dy) Py [ω(t) ∈ db]. (3.11) Recall from Definition 3.7 that δ(y) stands for the monopole and the dipole is represented by −δ0 (y). Therefore when a monopole starts from a point x, the distribution of the process at time t is Z dy δ(y − x) Py [ω(t) ∈ db] = p(t, b − x) db = Px [ω(t) ∈ db]. On the other hand, when a dipole starts from a point x, the distribution at time t is Z dy (−δ0 (y − x)) Py [ω(t) ∈ db] = ∂x p(t, x − b) db. We have a similar formula for the joint distribution of the first hitting time and place. When a monopole starts from a point x, the distribution is Z n dy δ(y − x) Py [τ0 (ω) ∈ dt, ω(τ0 ) is a monopole and in da] o (3.12) + Py [τ0 (ω) ∈ dt, ω(τ0 ) is a dipole and in da ] = same as the right hand side of (3.10). 9 On the other hand when a dipole starts from x, then the distribution is Z n dy (−δ0 (y − x)) Py [τ0 (ω) ∈ dt, ω(τ0 ) is a monopole and in da ] o + Py [τ0 (ω) ∈ dt, ω(τ0 ) is a dipole and in da ] n ∂K o ∂J = (t, x) δ(a) − (t, x) δ0 (a) dt da. ∂x ∂x (3.13) Remark 3.9. Informally speaking, (3.13) shows that a dipole generates both a monopole and a dipole at the hitting time of (−∞, 0), just like a monopole does. 3.4 Some Laplace transforms For future reference, we list some Laplace and Fourier-Laplace transforms of K(t, x), J(t, x) and p(t, x). Let λ > 0 and x ≥ 0. From Proposition 3.3, we have Z ∞ K̂(λ, x) ≡ dt exp{−λt} K(t, x) 0 ! (3.14) √ √ λ1/4 x π 1/4 √ − = 2 exp{−λ x/ 2} cos , 4 2 Z ˆ x) ≡ J(λ, ∞ dt exp{−λt} J(t, x) 0 √ √ 2 λ1/4 x = 1/4 exp{−λ1/4 x/ 2} sin √ , λ 2 Z ∞ Z ∞ Ĝ(λ, x, β) ≡ dt dy exp{−λt + iβy} Px [ω(t) ∈ dy] 0 −∞ exp{iβx} = . λ + β4 In addition, we define a new function Z ∞ Z 00 Ĝ (λ, x, β) ≡ dt exp{−λt + iβb}Px [ω(t) ∈ db]− 0 b∈R Z ∞ Z − dt exp{−λt + iβb}× 0 b∈R Z t Z n × Px [τ0 (ω) ∈ ds, ω(τ0 ) is a monopole and in da] s=0 a∈R o + Px [τ0 (ω) ∈ ds, ω(τ0 ) is a dipole and in da] Pa [ω(t − s) ∈ db] 1 ˆ(λ, x) . = Ĝ(λ, x, β) − K̂(λ, x) + iβ J λ + β4 (3.15) (3.16) (3.17) It is elementary to check that K̂(λ, 0) = 1, ∂x3 K̂(λ, 0) = ∂x K̂(λ, 0) = 0, √ 3/4 2λ , √ ∂x2 K̂(λ, 0) = − λ, ∂x4 K̂(λ, 0) = −λ, 10 ∂x5 K̂(λ, 0) = 0; (3.18) √ ˆ 0) = 0, ∂x J(λ, ˆ 0) = 1, ∂ 2 J(λ, ˆ 0) = − 2λ1/4 , J(λ, x √ ˆ 0) = 0, ∂ 5 J(λ, ˆ 0) = −λ; ∂ 3 Jˆ(λ, 0) = λ, ∂ 4 J(λ, x x (3.19) x Ĝ00 (λ, 0, β) = 0, ∂x Ĝ00 (λ, 0, β) = 0, √ √ −β 2 + λ + i 2λ1/4 β 2 00 ∂x Ĝ (λ, 0, β) = λ + β4 √ √ 3/4 − 2λ − iβ 3 − i λβ 3 00 ∂x Ĝ (λ, 0, β) = , λ + β4 ∂x4 Ĝ00 (λ, 0, β) = 1, 4 (3.20) ∂x5 Ĝ00 (λ, 0, β) = iβ. Killing boundaries for both particles Definition 4.1. Let P00 x [ω(t) ∈ db] = Px [ω(t) ∈ db]− Z t Z n − Px [τ0 (ω) ∈ ds, ω(τ0 ) is a monopole and in da] s=0 a∈R o + Px [τ0 (ω) ∈ ds, ω(τ0 ) is a dipole and in da] Pa [ω(t − s) ∈ db]. (4.1) We will refer to {P00 x [ω(t) ∈ db] : x ≥ 0} as a BP P with killing boundaries for both particles. Remark 4.2. Here is an intuitive description of BP P with killing boundaries for both particles. 1. A monopole starts from a point x ≥ 0, and moves according to the transition probability density p(t, x) until it hits the interval (−∞, 0). 2. After BP P hits (−∞, 0), both the monopole and dipole are killed. Theorem 4.3. (i) For x ≥ 0, b ≥ 0, and t > 0, we define a function p00 (t, x, b) ≡ p(t, b − x) − Then, we have Z Z ∞ ∞ dt 0 Z t 0 ds K(s, x) p(t − s, b) + J(s, x) ∂b p(t − s, b) . (4.2) db exp{−λt + iβb) p00 (t, x, b) = Ĝ00 (λ, x, β). −∞ (ii) The following linear functionals are identical and continuous on Bb [0, ∞), 00 P00 x [ω(t) ∈ db] = p (t, x, b) db. (4.3) Proof. We easily obtain (4.2) and (4.3) when we recall the definitions of the terms on the right hand side of (4.1) from Section 3.2. It is straightforward to check that (3.20) is the FourierLaplace transform of (4.2) in the usual sense. Part (ii) follows from the fact that p(t, x) ∈ S and from the estimates stated in Remark 3.4. 11 Remark 4.4. The following assertion easily follow from (4.2). (i) If f is a bounded continuous function on [0, ∞), then Z ∞ lim P00 x [ω(t) ∈ db] f (b) = f (x). t→0 0 (ii) If x = 0, P00 x [ω(t) ∈ A] = 0 for any Borel set A ∈ (0, ∞). As expected, the distribution P00 x [ω(t) ∈ db] solves (1.4) with the Dirichlet boundary condition. Theorem 4.5 ([14]). Let f be a bounded smooth function on the interval [0, ∞) such that f (0) = 0 = f 0 (0). We put Z ∞ v(t, x) ≡ P00 (4.4) x [ω(t) ∈ db] f (b). 0 Then in the classical sense, this v satisfies (1.4) with ` = 0 and m = 1. Corollary 4.6. P00 x [ω(t) ∈ db] satisfies the Chapman-Kolmogorov equations, that is Z ∞ db p00 (t, x, b) p00 (s, b, y) = p00 (t + s, x, y). (4.5) 0 Proof. Fix s > 0 and y ≥ 0. Theorem 4.5 asserts that Z ∞ v(t, x) ≡ db p00 (t, x, b) p00 (s, b, y) 0 is a solution of (1.4) where f (x) ≡ p00 (s, x, y), ` = 0 and m = 1. The function ṽ(t, x) ≡ p00 (t + s, x, y) is also a solution of (1.4). It is well known that the solution is unique so it follows that v = ṽ. The corollary is proved. Remark 4.7. Due to Corollary 4.6, we may consider P00 x [ω(t) ∈ db] of (4.3) as a finitely additive signed measure on cylinder sets in D[0, ∞) in the same way as in (2.4). 5 5.1 Reflecting boundary for the monopole Reflection for the monopole and killing of the dipole The construction of the process BP P with the boundary conditions specified above will use an approximating sequence. We will prove existence of a family { Pr0 x [ω(t) ∈ db] : x ≥ 0} satisfying 0 Pr0 x [ω(t) ∈ db] = Px [ω(t) ∈ db]+ Z tZ ∞ + Px [τ0 (ω) ∈ ds, ω(τ0 ) is a monopole and in da] Pr0 +a [ω(t − s) ∈ db]. 0 (5.1) −∞ Remark 5.1. Heuristically, { Pr0 x [ω(t) ∈ db] : x ≥ 0} represents the following process. Fix > 0. 12 1. A monopole starts from a point x ≥ 0, and moves according to the transition probability density p(t, x) until it hits the interval (−∞, 0). 2. If ω(τ0 ) is a monopole when BP P hits (−∞, 0), then it restarts from the point + ω(τ0 ). 3. If ω(τ0 ) is a dipole when BP P hits (−∞, 0), then it is killed. Assume for the moment that Pr0 x [ω(t) ∈ db] is a Schwartz temperate distribution and denote its Fourier-Laplace transform by Z ∞ Z r0 r0 U (λ, x, β) ≡ dt Px [ω(t) ∈ db] exp{−λt + iβb}]. 0 Apply the Fourier-Laplace transform to the both sides of (5.1) to obtain U r0 (λ, x, β) = Ĝ00 (λ, x, β) + K̂(λ, x) U r0 (λ, , β), x > 0. (5.2) Lemma 5.2. (i) When > 0, the following is a solution of (5.2): U r0 (λ, x, β) = Ĝ00 (λ, x, β) + K̂(λ, x) Ĝ00 (λ, , β) 1 − K̂(λ, ) . (5.3) (ii) For small > 0, this U r0 satisfies (3.4) for any c > 0, and it is the Fourier-Laplace transform of a Schwartz temperate distribution Pr0 x [ω(t) ∈ db]. Proof. (i) Put x = in (5.2) to see that U r0 (λ, , β) = Ĝ00 (λ, , β) + K̂(λ, ) U r0 (λ, , β). This linear equation is easily solved with respect to U r0 . This and (5.2) yield (5.3). (ii) Let <λ > c > 0. We take the principal value for λ1/4 and recall that −π/8 < arg λ1/4 < π/8. After applying this fact to (3.14) through (3.17), we see that (3.4) holds for K̂, Jˆ, and Ĝ00 , and also for the fraction on the right hand side of (5.3) when > 0 is small. Now Lemma 3.1 implies our assertion. Lemma 5.3. In the distribution sense, Pr0 x [ω(t) ∈ db] converges to a limit as → 0. Definition 5.4. This above limit will be denoted by {Pr0 x [ω(t) ∈ db] : x ≥ 0} and called the distribution of BP P with the reflecting boundary for the monopole and the killing boundary for the dipole. Proof of Lemma 5.3. Recalling (3.18) through (3.20), we let → 0 in (5.3), and obtain U r0 (λ, x, β) ≡ lim U r0 (λ, x, β) →0 = Ĝ00 (λ, x, β) − K̂(λ, x) ∂x2 Ĝ00 (λ, 0, β) ∂x2 K̂(λ, 0) (5.4) . Now U r0 clearly satisfies (3.4), and so it is the Fourier-Laplace transform of a Schwartz temperate distribution, that is Pr0 x [ω(t) ∈ db]. 13 Assuming that <λ > 0, we calculate the usual inverse Fourier-Laplace transform of ∂x2 Ĝ00 (λ, 0, β)/∂x2 K̂(λ, 0). Let b ≥ 0. By the residue theorem, we have Z L ∂ 2 Ĝ00 (λ, 0, β) 1 r0 I (λ, b) ≡ dβ exp{−iβb} − x lim 2π L→∞ −L ∂x2 K̂(λ, 0) (5.5) i 1/4 1/4 =√ exp{λ θ2 b} − exp{λ θ2 b} , 2λ3/4 where θ2 ≡ exp{i3π/4} and θ2 is its complex conjugate. Applying the usual inverse Laplace transform to this I r0 , we obtain Z 1+iM 1 r0 Q (t, b) ≡ dλ exp{λt} I r0 (λ, b) lim 2πi M →∞ 1−iM (5.6) Z 2 ∞ −y 4 t = cos yb + sin yb − exp{−yb} , dy e π 0 which is a smooth bounded function in b ≥ 0 if t > 0. Theorem 5.5. For x ≥ 0, b ≥ 0, and t > 0, we define a function Z t pr0 (t, x, b) ≡ p00 (t, x, b) + ds K(s, x) Qr0 (t − s, b), (5.7) 0 where p00 is the function in (4.2) and Qr0 is defined in (5.6). Then we have r0 Pr0 x [ω(t) ∈ db] = p (t, x, b) db, (5.8) both linear functionals in (5.8) are continuous on Bb [0, ∞) and Z ∞ Pr0 x [ω(t) ∈ db] = 1. 0 Proof. The first part of the theorem follows from (5.4), (5.6), and Lemma 5.3. From (5.4) with (3.18) through (3.20), we deduce that U r0 (λ, x, 0) = 1/λ, and this justifies the last assertion. The distribution Pr0 x is related to the PDE in (1.4) with the Neumann boundary conditions. Theorem 5.6. (i) Let f be a bounded smooth function on the interval [0, ∞) such that f 0 (0) = 0 = f 00 (0). We put Z ∞ v(t, x) ≡ Pr0 (5.9) x [ω(t) ∈ db] f (b). 0 Then in the classical sense, this v satisfies (1.4) with ` = 1 and m = 2. (ii) Pr0 x [ω(t) ∈ db] satisfies Chapman-Kolmogorov equations (4.5), and Remark 4.7 holds with r0 Px [ω(t) ∈ db] instead of P00 x [ω(t) ∈ db]. Proof. Recall (3.18) through (3.20). From (5.4), we have ∂x4 U r0 (λ, x, β) + λU r0 (λ, x, β) = exp{iβ x} for ∀x > 0, and ∂x U r0 (λ, 0, β) = 0 (5.10) and ∂x2 U r0 (λ, 0, β) = 0. Since the Fourier-Laplace transform is unique for our pr0 , these prove the first part of the theorem. The second part is immediate from the same argument as in the proof of Corollary 4.6. 14 5.2 An analogy to the classical reflecting Brownian motion For the Brownian motion B(t), it is well known that the following two process X1 and X2 have the same distributions. They are referred to as reflecting Brownian motions. Let τ0 be the first hitting time of the point x = 0. X1 (t) ≡ |B(t)|, X2 (t) ≡ B(t) B(t) − inf τ0 ≤s≤t B(s) if t < τ0 , if t ≥ τ0 . It is not difficult to check that similarly defined transformations of BP P do not have identical distributions. Keeping this fact in mind, we define a new BP P as follows. Definition 5.7. Let {Pm x [ω(t) ∈ db] : x ≥ 0} be defined by 00 Pm x [ω(t) ∈ db] = Px [ω(t) ∈ db]+ Z tZ ∞ + Px [τ0 (ω) ∈ ds, ω(τ0 ) is a monopole and in da]× 0 −∞ (5.11) × Pa [ω(t − s) − inf ω(t) ∈ db]. u≤t−s Remark 5.8. We present an intuitive view of the last definition. 1. A monopole starts from a point x ≥ 0, and moves according to the transition probability density p(t, x) until it hits the interval (−∞, 0). 2. If ω(τ0 ) is a dipole when BP P hits (−∞, 0), then it is killed. 3. If ω(τ0 ) is a monopole when BP P hits (−∞, 0), then it restarts from the point x = 0, and moves according to the distribution of [ω(t + τ0 ) − inf τ0 ≤s≤t+τ0 ω(s)]. In [14], the following was proved. For b ≥ 0, P0 [ω(t) − inf ω(s) ∈ db] = Qr0 (t, b) db, 0≤s≤t where Qr0 is given in (5.6). We record this partial analogy with the classical reflected Brownian motion in the next theorem. m Theorem 5.9. The distributions Pr0 x [ω(t) ∈ db] in Theorem 5.5 and Px [ω(t) ∈ db] in (5.11) are identical. Proof. The result follows easily from (5.11) and Theorem 5.5. 5.3 Reflecting boundaries for both particles The construction of BP P with these boundary conditions requires an approximation procedure similar to that in the previous section. We will prove that there exists a family { Prr x [ω(t) ∈ 15 db] : x ≥ 0} satisfying the following equation. 00 Prr x [ω(t) ∈ db] = Px [ω(t) ∈ db]+ Z t Z ∞ n + Px [τ0 (ω) ∈ ds, ω(τ0 ) is a monopole and in da] s=0 a=−∞ o + Px [τ0 (ω) ∈ ds, ω(τ0 ) is a dipole and in da ] Prr +a [ω(t − s) ∈ db]. (5.12) Remark 5.10. Heuristically, we are dealing with the following process. Let be a fixed positive number. 1. A monopole starts from a point x ≥ 0, and moves according to the transition probability density p(t, x) until it hits the interval (−∞, 0). 2. After BP P hits (−∞, 0), it restarts from the point + ω(τ0 ) irrespective of whether ω(τ0 ) is a monopole or a dipole. rr its Fourier-Laplace Assuming for the moment existence of Prr x [ω(t) ∈ db], we denote by U transform. Recall (3.13) and note that (5.12) is transformed into the following equation. For x > 0, U rr (λ, x, β) = Ĝ00 (λ, x, β)+ + K̂(λ, x) U rr (λ, , β) + Jˆ(λ, x) ∂x U rr (λ, , β). (5.13) Lemma 5.11. (i) For > 0, the following is a solution of (5.13): U rr (λ, x, β) = Ĝ00 (λ, x, β) ˆ λ)) Ĝ00 (λ, , β) + J(λ, ˆ ) ∂x Ĝ00 (λ, , β) (1 − ∂x J(, + K̂(λ, x) ˆ )) − ∂x K̂(, λ) Jˆ(λ, ) (1 − K̂(λ, ))(1 − ∂x J(λ, (5.14) 00 00 ˆ x) ∂x K̂(λ, ) Ĝ (λ, , β) − (1 − K̂(λ, )) ∂x Ĝ (λ, , β) . + J(λ, ˆ )) − ∂x K̂(λ, ) Jˆ(λ, ) (1 − K̂(λ, ))(1 − ∂x J(λ, (ii) For small > 0, this U rr satisfies (3.4) for any c > 0, and it is the Fourier-Laplace transform of a Schwartz temperate distribution Prr x [ω(t) ∈ db]. Proof. (i) Substitute for x in (5.13). Then differentiate both sides of (5.13) with respect to x and then again substitute for x. As a result, we obtain the following equations, U rr (λ, , β) = Ĝ00 (λ, , β) + K̂(λ, ) U rr (λ, , β) ˆ ) ∂x U rr (λ, , β), +J(λ, ∂x U rr (λ, , β) = ∂x Ĝ00 (λ, , β) + ∂x K̂(λ, ) U rr (λ, , β) +∂x Jˆ(λ, ) ∂x U rr (λ, , β). We can solve the equations for U rr (λ, , β) and ∂x U rr (λ, , β) and substitute the results into (5.13). Then (5.14) easily follows. Part (ii) can be proved the same way as part (ii) of Lemma 5.2. 16 Theorem 5.12. In the sense of convergence of distributions, Prr x [ω(t) ∈ db] converges to Pr0 [ω(t) ∈ db] defined in Theorem 5.5, as → 0. x Proof. If we let → 0 in (5.14) then we obtain lim U rr (λ, x, β) = Ĝ0 (λ, x, β) − K̂(λ, x) →0 ∂x2 Ĝ0 (λ, 0, β) ∂x2 K̂(λ, 0) . The right hand side equals to U r0 in the proof of Lemma 5.4, so it is the Fourier-Laplace transform of Pr0 x [ω(t) ∈ db]. 6 Killing of monopole and reflection for dipole Once again, we start with an approximating sequence. We will find a solution { P0r x [ω(t) ∈ db] : x ≥ 0, } to the equation 00 P0r x [ω(t) ∈ db] = Px [ω(t) ∈ db]+ Z tZ ∞ + Px [τ0 (ω) ∈ ds, ω(τ0 ) is a dipole and in da] P0r +a [ω(t − s) ∈ db]. 0 (6.1) −∞ Remark 6.1. Intuitively speaking, we have the following model. Consider fixed > 0. 1. A monopole starts from a point x ≥ 0, and moves according to the transition probability density p(t, x) until it hits the interval (−∞, 0). 2. If ω(τ0 ) is a monopole when BP P hits (−∞, 0), then it is killed. 3. If ω(τ0 ) is a dipole when BP P hits (−∞, 0), then it restarts from the point + ω(τ0 ). Suppose { P0r x [ω(t) ∈ db] : x ≥ 0, } solving (6.1) exists. We denote the Fourier-Laplace transform 0r of Px [ω(t) ∈ db] by U 0r (λ, x, β). Then we apply the Fourier-Laplace transform to (6.1). Using (3.13), we obtain for x > 0, ˆ x) ∂x U r0 (λ, , β). U 0r (λ, x, β) = Ĝ00 (λ, x, β) + J(λ, (6.2) The following lemma can be proved the same way as Lemma 5.11 so we omit its proof. Lemma 6.2. (i) For > 0, the following is a solution of (6.2): 0 ˆ x) ∂x Ĝ (λ, , β) . U 0r (λ, x, β) = Ĝ00 (λ, x, β) + J(λ, ˆ ) 1 − ∂x J(λ, (6.3) (ii) For small > 0, this U 0r satisfies (3.4) with any c > 0, and it is the Fourier-Laplace transform of a Schwartz temperate distribution Por x [ω(t) ∈ db]. Lemma 6.3. In the sense of convergence of distributions, P0r x [ω(t) ∈ db] converges to a limit as → 0. 17 Definition 6.4. We denote this limit by {P0r x [ω(t) ∈ db] : x ≥ 0}, and call it the distribution of BP P with the killing boundary for the monopole and the reflecting boundary for the dipole. Proof of Lemma 6.3. We let tend to 0 in (6.3). Then (3.19) and (3.20) imply that U 0r (λ, x, β) ≡ lim U 0r (λ, x, β) →∞ 2 00 ˆ x) ∂x Ĝ (λ, 0, β) . = Ĝ00 (λ, x, β) − J(λ, ˆ 0) ∂x2 J(λ, (6.4) When we take the principal value of λ1/4 , U 0r satisfies (3.4), and Lemma 3.1 implies our assertion. ˆ 0) admits the usual inverse Fourier–Laplace transform. We If <λ > 0, ∂x2 Ĝ00 (λ, 0, β)/∂x2 J(λ, present the calculations. Let b ≥ 0. The residue theorem implies that Z L ∂ 2 Ĝ00 λ, 0, β) 1 0r I (λ, b) ≡ dβ exp{−iβb} − x lim ˆ 0) 2π L→∞ −L ∂x2 J(λ, (6.5) i 1/4 1/4 = √ exp{λ b θ2 } − exp{λ b θ2 } . 2 λ For t > 0, we obtain that Z 1+iM 1 0r Q (t, b) ≡ − dλ exp{λt} I 0r (λ, b) lim 2πi M →∞ 1−iM (6.6) Z 2 ∞ 4 = dy exp{−y t} y sin(yb). π 0 One can easily check that the density Q0r is a smooth bounded function of non-negative b if t > 0, and it is integrable in t. Theorem 6.5. For x ≥ 0, b ≥ 0, and t > 0, we define a function Z t p0r (t, x, b) ≡ p00 (t, x, b) + ds J(s, x) Q0r (t − s, b), (6.7) 0 where Q0r is the function in (6.6). Let {P0r x [ω(t) ∈ db] : x ≥ 0} denote the distribution in Definition 6.4. Then we have 0r P0r x [ω(t) ∈ db] = p (t, x, b) db, (6.8) and these distributions are continuous linear functionals on Bb [0, ∞). Proof. The theorem is immediate from Lemma 6.2 and (6.6). The following theorem is easily proved using (6.4) in the same way as Theorem 5.11 so we omit its proof. Theorem 6.6. (i) Let f be a bounded smooth function on the interval [0, ∞) such that f (0) = 0 = f 00 (0). We put Z ∞ v(t, x) ≡ P0r (6.9) x [ω(t) ∈ db] f (b). 0 Then in the classical sense, this v satisfies (1.4) with ` = 0 and m = 2. (ii) P0r x [ω(t) ∈ db] satisfies Chapman-Kolmogorov equations (4.5), and the assertion in Remark 0 4.7 holds with P0r x [w(t) ∈ db] in place of Px [ω(t) ∈ db]. 18 7 Stopped monopole 7.1 Stopped monopole and dipole Definition 7.1. Let {Pss x [ω(t) ∈ db] : x ≥ 0} be defined by 00 Pss x [ω(t) ∈ db] = Px [ω(t) ∈ db]+ Z tZ ∞n + Px [τ0 (ω) ∈ ds, ω(τ0 ) is a monopole and in da] 0 −∞ o + Px [τ0 (ω) ∈ ds, ω(τ0 ) is a dipole and in da] δ(b − a) db. (7.1) Remark 7.2. The family of distributions given in the last definition represents a BP P whose both particles are stopped after exiting (0, ∞). An intuitive description of the process is the following. 1. A monopole starts from a point x ≥ 0, and moves according to the transition probability density p(t, x) until it hits the interval (−∞, 0). 2. After BP P hits the interval (−∞, 0), both particles are stopped. Theorem 7.3. For x ≥ 0, b ≥ 0, and t > 0, we define a linear functional on C1b [0, ∞), Z t ss 00 p (t, x, b) ≡ p (t, x, b) + ds K(s, x) δ(b) − J(s, x) δ0 (b) . (7.2) 0 Then we have ss Pss x [ω(t) ∈ db] = p (t, x, b) db. Both sides of the last formula are continuous linear functionals on Proof. The result follows easily from (3.13) and Definition 7.1. (7.3) C1b [0, ∞). Theorem 7.4. (i) Let f be a bounded smooth function defined on the interval [0, ∞) such that f (4) (0) = 0 = f (5) (0). We define Z ∞ v(t, x) ≡ Pss (7.4) x [ω(t) ∈ db] f (b). 0 Then in the classical sense, this v satisfies (1.4) with ` = 4 and m = 5. (ii) Pss x [ω(t) ∈ db] satisfies Chapman-Kolmogorov equations (4.5) in the sense of a linear functional on C1b [0, ∞). ss Proof. We apply the Fourier-Laplace transform to Pss x [ω(t) ∈ db], and denote it by U (λ, x, β). Then (7.2) is transformed into 1 ˆ x) iβ , + J(λ, (7.5) λ λ from which we see that (5.10) holds for this U ss . Recalling (3.18), (3.19) and (3.20), we easily see that ∂x4 U ss (λ, 0, β) = 0 and ∂x5 U ss (λ, 0, β) = 0. U ss (λ, x, β) = Ĝ00 (λ, x, β) + K̂(λ, x) This proves part (i) of the theorem, since the Fourier-Laplace transform is unique. For (ii), it is sufficient to note that ∂x pss (t, x, b) is a linear functional on C1b [0, ∞) and the left hand side of (4.5) is well-defined with p00 (t, x, b) replaced by ∂x pss (t, x, b). 19 7.2 Stopped monopole and killed dipole Definition 7.5. Let {Ps0 x [ω(t) ∈ db] : x ≥ 0} be given by 00 Ps0 x [ω(t) ∈ db] = Px [ω(t) ∈ db]+ Z t Z ∞ + Px [τ0 (ω) ∈ ds, ω(τ0 ) is a monopole and in da] δ(a − b) db. s=0 (7.6) a=−∞ Remark 7.6. A heuristic view of BP P with the boundary behavior specified in the section title is this. 1. A monopole starts from a point x ≥ 0, and moves according to the transition probability density p(t, x) until it hits the interval (−∞, 0). 2. If ω(τ0 ) is a monopole when BP P hits (−∞, 0), then it is stopped at the point ω(τ0 ). 3. If ω(τ0 ) is a dipole when BP P hits (−∞, 0), then it is killed. Theorem 7.7. For x ≥ 0, b ≥ 0, and t > 0, we define a linear functional on Bb [0, ∞), Z t s0 0 p (t, x, b) ≡ p (t, x, b) + ds K(s, x) δ(b). (7.7) 0 We have s0 Ps0 x [ω(t) ∈ db] = p (t, x, b) db, (7.8) where both expressions are continuous linear functionals on Bb [0, ∞). Proof. The result follows easily from the definition (7.6). Theorem 7.8. (i) Let f be a bounded smooth function on the interval [0, ∞) such that f 0 (0) = 0 = f (4) (0). We put Z ∞ v(t, x) ≡ Ps0 (7.9) x [ω(t) ∈ db] f (b). 0 Then in the classical sense, this v satisfies (1.4) with ` = 1 and m = 4. (ii) Ps0 x [ω(t) ∈ db] satisfies Chapman-Kolmogorov equations (4.5) in the sense of a linear functional on Bb [0, ∞). Proof. Let U s0 (λ, x, β) be the Fourier-Laplace transform of Ps0 x . Apply the Fourier-Laplace transform to (7.7). The equation is transformed to U s0 (λ, x, β) = Ĝ00 (λ, x, β) + K̂(λ, x) 1 . λ (7.10) From this, we see that (5.10) holds for this U s0 and that ∂x U s0 (λ, 0, β) = 0 and ∂x4 U s0 (λ, 0, β) = 0. The first assertion of the theorem follows from this and from uniqueness of the Fourier-Laplace transform. The proof of the second is the same as in Theorem 7.3. We present an example of a “not well-posed problem” for (1.4): 20 Example 7.9. Let f be a bounded smooth function on the interval [0, ∞) such that f 0 (0) = 0, f 00 (0) = 0, and f (4) (0) = 0. Recalling Theorems 5.7 and 7.8, we define functions Z ∞ Z ∞ v1 (t, x) ≡ Pr0 [ω(t) ∈ db] f (b); v2 (t, x) ≡ Ps0 [ω(t) ∈ db] f (b). o o These v1 and v2 are classical solutions of (1.4) with ` = 1. We consider their Laplace transforms Z ∞ Vj (λ, x) ≡ dt exp{−λt} vj (t, x), j = 1, 2, 0 which satisfy ∂x4 Vj (λ, x) = −λVj (λ, x) + f (x), j = 1, 2. Since they are smooth and both satisfy ∂x Vj (λ, 0) = 0, we have 0 = ∂x5 Vj (λ, 0) = −λ∂x Vj (λ, 0) + f 0 (0) = 0, j = 1, 2. This fact implies that solutions to (1.4) are not unique when ` = 1 and m = 5. Note that an analogous argument also holds when ` = 0 and m = 4. 7.3 Stopped monopole and reflected dipole This case calls for an approximation procedure similar to those discussed in earlier sections. We are looking for a family { Psr x [ω(t) ∈ db] : x ≥ 0} satisfying 00 Psr x [ω(t) ∈ db] = Px [ω(t) ∈ db]+ Z tZ ∞ + Px [τ0 (ω) ∈ ds, ω(τ0 ) is a monopole and in da] δ(a − b) db 0 −∞ ∞ Z tZ + 0 −∞ (7.11) Px [τ0 (ω) ∈ ds, ω(τ0 ) is a dipole and in da] Psr +a [ω(t − s) ∈ db]. Remark 7.10. An intuitive description of { Psr x [ω(t) ∈ db] : x ≥ 0} is the following. Let be a fixed positive number. 1. A monopole starts from a point x ≥ 0, and moves according to the transition probability density p(t, x) until it hits the interval (−∞, 0). 2. If ω(τ0 ) is a monopole when BP P hits (−∞, 0), then it is stopped at the point ω(τ0 ). 3. If ω(τ0 ) is a dipole when BP P hits (−∞, 0), then it restarts from the point + ω(τ0 ). Let U sr (λ, x, β) denote the Fourier-Laplace transform of Psr x , if it exists. Recall (3.13) and note that (7.11) can be transformed to U sr (λ, x, β) = Ĝ00 (λ, x, β)+ 1 ˆ x) ∂x U sr (λ, , β). + K̂(λ, x) + J(λ, λ (7.12) The following lemma is similar to several lemmas presented earlier in the paper so we omit its proof. 21 Lemma 7.11. (i) For > 0, the following is a solution of (7.12): 1 U sr (λ, x, β) = Ĝ00 (λ, x, β) + +K̂(λ, ) + λ 00 ˆ ) ∂x Ĝ (λ, , β) + ∂x K̂(λ, )/λ . + J(λ, ˆ ) 1 − ∂x J(λ, (7.13) (ii) When > 0 is small, this U sr satisfies (3.4) for every c > 0 and it is the Fourier-Laplace transform of a Schwartz temperate distribution Psr x [ω(t) ∈ db]. Lemma 7.12. Psr x [ω(t) ∈ db] converges to a limit as → 0, in the distribution sense. Definition 7.13. The above limit will be denoted by {Psr x [ω(t) ∈ db] : x ≥ 0}, and called the distribution of BP P with stopped monopole and reflecting dipole on the boundary. Proof. Recall (3.18) through (3.20), and let → 0 in (7.12). Then 1 U sr (λ, x, β) ≡ lim U sr (λ, x, β) = Ĝ00 (λ, x, β) + K̂(λ, x) − →0 λ 2 00 2 ˆ x) ∂x Ĝ (λ, 0, β) + ∂x K̂(λ, 0)/λ . − J(λ, ˆ 0) ∂x2 J(λ, This satisfies (3.4), so Lemma 3.1 implies our assertion. (7.14) We will apply the inverse Fourier-Laplace transform to − ∂x2 Ĝ00 (λ, 0, β) + ∂x2 K̂(λ, 0)/λ ∂ 2 Ĝ00 (λ, 0, β) ∂ 2 K̂(λ, 0) =− x − x ˆ 0) ˆ 0) ∂x2 J(λ, ∂x2 J(λ, λ ∂x2 Jˆ(λ, 0) (7.15) in the distribution sense. We already know a formula for the first term of (7.15), which is the same as Q0r in (6.6). As for the second term − ∂x2 K̂(λ, 0) 1 , =√ 2 ˆ 2 λ3/4 λ ∂x J(λ, 0) we can apply the inverse Fourier-Laplace transform in the sense of Schwartz temperate distribution, to obtain Z 2 ∞ − dy exp{−y 4 t} δ(b). π 0 We conclude that the inverse Fourier-Laplace transform of (7.15) is Z 2 ∞ sr 0r Q (t, b) ≡ Q (t, b) − dy exp{−y 4 t} δ(b). π 0 (7.16) This is a linear functional on Bb [0, ∞). The following theorem is immediate from Lemma 7.12 and the above arguments. Therefore we omit its proof. 22 Theorem 7.14. For x ≥ 0, b ≥ 0, and t > 0, we define a linear functional on Bb [0, ∞), sr Z 00 t p (t, x, b) ≡ p (t, x, b) + ds K(s, x) δ(b)+ 0 Z s + ds J(s, x) Qsr (t − s, b) (7.17) 0 where Qsr is given in (7.16). We have sr Psr x [ω(t) ∈ db] = p (t, x, b) db. (7.18) Both sides are continuous linear functionals on Bb [0, ∞). Theorem 7.15. (i) Let f be a bounded smooth function on the interval [0, ∞) such that f 00 (0) = 0 = f (4) (0). We put Z v(t, x) ≡ ∞ 0 Psr x [ω(t) ∈ db] f (b). (7.19) Then in the classical sense, this v satisfies (1.4) with ` = 2 and m = 4. (ii) Psr x [ω(t) ∈ db] satisfies Chapman-Kolmogorov equations (4.5), and Remark 4.7 holds with sr Px [ω(t) ∈ db] in place of P0x [ω(t) ∈ db]. Proof. Note that the Fourier-Laplace transform is unique. By (7.14), we see that (5.10) holds for U sr . Moreover using (3.18) through (3.20), we have ∂x2 U sr (λ, 0, β) = 0 and ∂x4 U sr (λ, 0, β) = 0. These prove (i), and (ii) can be proved by the same arguments as similar results in earlier sections. 8 Stopped dipole 8.1 Stopped dipole and killed monopole Definition 8.1. Let {P0s x [ω(t) ∈ db] : x ≥ 0} be given by 00 P0s x [ω(t) ∈ db] = Px [ω(t) ∈ db]+ Z tZ ∞ + Px [τ0 (ω) ∈ ds, ω(τ0 ) is a dipole and in da]δ(a − b) db. 0 (8.1) ∞ Remark 8.2. Intuitively, we are dealing with the following model. 1. A monopole starts from a point x ≥ 0, and moves according to the transition probability density p(t, x) until it hits the interval (−∞, 0). 2. If ω(τ0 ) is a monopole when BP P hits (−∞, 0), then it is killed. 3. If ω(τ0 ) is a dipole when BP P hits (−∞, 0), then it is stopped at the point ω(τ0 ). 23 The following theorem can be easily derived from (3.13). Theorem 8.3. For x ≥ 0, b ≥ 0, and t > 0, we define a linear functional on C1b [0, ∞), Z s 0s 00 p (t, x, b) ≡ p (t, x, b) + ds J(s, x) δ0 (b). (8.2) 0 Then we have 0s P0s x [ω(t) ∈ db] = p (t, x, b) db, (8.3) where both expressions are continuous linear functionals on C1b [0, ∞). Theorem 8.4. (i) Let f be a bounded smooth function on the interval [0, ∞) such that f (0) = 0 = f (5) (0). We put Z ∞ v(t, x) ≡ P0s (8.4) x [ω(t) ∈ db] f (b). 0 Then in the classical sense, this v satisfies (1.4) with ` = 0 and m = 5. (ii) P0s x satisfies Chapman-Kolmogorov equations (4.5). Proof. If we denote by U 0s (x, λ, β) the Fourier-Laplace transform of P0s x , then (8.3) is transformed into the following equality: ˆ λ, β) U 0s (x, λ, β) = Ĝ00 (x, λ, β) + J(x, iβ , λ (8.5) from which we see that (5.10) holds for this U 0s . Using (3.18) through (3.20), we have U 0s (λ, 0, β) = 0 and ∂x5 U 0s (λ, 0, β) = 0. These prove (i). Note that ∂x p0s (t, x, b) is a linear functional on C1b [0, ∞). The second part of the theorem can be proved by arguments used earlier in the paper. 8.2 Stopped dipole and reflected monopole For the last time in this paper we use an approximation procedure to construct a process. The first step is to find { Prs x [ω(t) ∈ db] : x ≥ 0} such that 0 Prs x [ω(t) ∈ db] = Px [ω(t) ∈ db]+ Z tZ ∞ + Px [τ0 (ω) ∈ ds, ω(τ0 ) is a monopole and in da] Prs +ω(τ0 ) [ω(t) ∈ db]+ 0 −∞ ∞ Z tZ + 0 −∞ (8.6) Px [τ0 (ω) ∈ ds, ω(τ0 ) is a dipole and in da] δ(b − a)db. Remark 8.5. The approximating process evolves according to the following rules. Fix > 0. 1. A monopole starts from a point x ≥ 0, and moves according to the transition probability density p(t, x) until it hits the interval (−∞, 0). 2. If ω(τ0 ) is a monopole when BP P hits (−∞, 0), then it restarts from the point + ω(τ0 ). 24 3. If ω(τ0 ) is a dipole when BP P hits (−∞, 0), then it is stopped at the point ω(τ0 ). Assuming existence, we denote by U rs (λ, x, β) the Fourier-Laplace transform of Prs x . Recall (3.13). We see that (8.6) is transformed into the following equation: U rs (λ, x, β) = Ĝ00 (λ, x, β) + K̂(λ, x) U sr (λ, , β) + Jˆ(λ, x) iβ . λ (8.7) We omit the proof of the following result. Lemma 8.6. (i) For > 0, the following is a solution of (8.7): ˆ x) iβ + U rs (λ, x, β) = Ĝ00 (λ, x, β) + J(λ, λ 00 ˆ )/λ Ĝ (λ, , β) + iβ J(λ, + K̂(λ, x) . 1 − K̂(λ, ) (8.8) (ii) When > 0 is small, this U rs satisfies (3.4) for every c > 0 and it is the Fourier-Laplace transform of a Schwartz temperate distribution Prs x [ω(t) ∈ db]. We have just proved that if > 0, then there exist distributions Prs x [ω(t) ∈ db] satisfying equations (8.6). However the following proposition asserts that we cannot pass to 0 with . Hence we cannot construct a BP P with reflection for the monopole and stopped dipole, at least not using our method. Proposition 8.7. (8.8) diverges as → 0. Remark 8.8. Proposition 8.7 is not very surprising if we consider the corresponding problem (1.4). Table 1.2 suggests that, if a BP P with reflection for the monopole and stopped dipole exists, then it corresponds to (1.4) with ` = 1 and m = 5. But solutions to this boundary value problem (1.4) are not unique. So we cannot find a “fundamental solution” of such (1.4), which would serve as a transition density for the corresponding BP P . Proof of Proposition 8.7. From (3.18) through (3.20), we see that 2 2 ∂x K̂(λ, 0) + o(2 ) 2 √ 2 λ =− + o(2 ), 2 1 − K̂(λ, ) = Ĝ00 (λ, , β) + This shows that iβ Jˆ(λ, ) iβ ˆ 0) + o() = × ∂x J(λ, λ λ iβ = + o(). λ Ĝ00 (λ, , β) + iβ Jˆ(λ, )/λ 1 − K̂(λ, ) diverges as → 0, and so (8.8) also diverges. 25 (8.9) (8.10) 9 Conservation of charge Let {Qx [ω(t)] : x ≥ 0} be the generic notation for the distribution of BP P on [0, ∞) with any boundary conditions for the monopole and dipole. Definition 9.1. We say that Qx [ω(t) ∈ db] is conservative if Z ρ(t, x) ≡ Qx [ω(t) ∈ db] = 1 for all t > 0 and x > 0. (9.1) [0,∞) Theorem 9.2. Let {Qx [ω(t) ∈ db] : x ≥ 0} be the distribution of any BP P on [0, ∞) constructed in this paper. Then Qx [ω(t) ∈ db] is conservative if and only if there is no killing for the monopole. Remark 9.3. This result intuitively supports for the names “monopole” and “dipole.” Since a dipole carries two charges of equal magnitude but opposite signs, its loss does not change the total amount of charge. On the other hand, the monopole carries charge of single sign, and its loss changes the total amount of charge. Proof of Theorem 9.2. Inspecting Table 1.2, we see that there is no killing for the monopole if and only if ` ≥ 1 and m ≥ 1. Assume that ` ≥ 1 and m ≥ 1 in (1.4). Since ρ(0, x) = 1 in view of (9.1), we see that ρ is a solution of (1.4) with the constant initial function f (x) ≡ 1. The constant function v(t, x) ≡ 1 is a solution of (1.4) with f ≡ 1, if ` ≥ 1 and m ≥ 1. Now uniqueness of the solution implies that ρ(t, x) ≡ 1, and Qx [ω(t) ∈ db] is conservative. On the other hand, if ` = 0, then the constant v(t, x) ≡ 1 is not a solution of (1.4) with the constant initial function f ≡ 1, and uniqueness of the solution implies that ρ(t, x) 6= 1 for some (t, x). In this case, Qx [ω(t) ∈ db] is not conservative. References [1] Beghin, L., Hochberg, K., and Orsingher, E., Conditional maximal distributions of processes related to higher-order heat-type equations, Stoch. Proc. Appl., 85 (2000), 209–223. [2] Burdzy, K., Some path properties of iterated Brownian motion, in Seminar on Stochastic Process 1992, E. Çinlar, K. L. Chung, and M. Sharpe, eds., Birkhaüser–Boston (1993), 67–83; Math. Review 95c:60075. [3] Burdzy, K. and Ma̧drecki, A., An asymptotically 4-stable process, J. Fourier Anal. Appl. Special Issue, Proceeding of the Conference in Honor of J. P. Kahane, 1995, 97-117; Math. Review 96j:60072. [4] Burdzy, K. and Ma̧drecki, A., Ito formula for an asymptotically 4-stable process, Ann. of Appl. Prob., 6 (1996), 200-217; Math. Review 99a:60042. [5] Doetsch, G., Introduction to the Theory and Application of the Laplace Transformation, Springer– Verlag, 1974; Math. Review 49#9549. 26 [6] Eidelman, S. D. and Zhitarashu, N. V., Parabolic Boundary Value Problems, Oper. Theor. Adv. Appl., Vol. 101, Birkhäuser, 1998; Math. Review 94m:35134. [7] Funaki, T., Probabilistic construction of the solution of some higher order parabolic differential equations, Proc. Japan Acad., 55 (1979), 176–179; Math. Review 80h:60075. [8] Hochberg, K. J., A signed measure on path space related to Wiener measure, Ann. Prob., 6 (1978), 433–458; Math. Review 80i:60111. [9] Krylov, Y. Ju., Some properties of the distribution corresponding to the equation ∂u/∂t = (−1)q+1 ∂ 2q u/∂ 2q x, Soviet Math. Dokl., 1 (1960), 760–763; Math. Review 22 #9722. [10] Nakajima, T. and Sato, S., On the joint distribution of the first hitting time and the first hitting place to the space-time wedge domain of a biharmonic pseudo process, Tokyo J. Math., 22 (1999), 399–413; Math. Review 2000k:60072. [11] Nishioka, K., Stochastic calculus for a class of evolution equations, Japan. J. Math., 11 (1985), 59–102; Math. Review 88f:60101. [12] Nishioka, K., A stochastic solution of a high order parabolic equation, J. Math. Soc. Japan., 39 (1987), 209–231; Math. Review 89f:35097. [13] Nishioka, K., Monopole and dipole of a biharmonic pseudo process, Proc. Japan Acad., Ser. A 72 (1996), 47–50; Math. Review 97g:60117. [14] Nishioka, K., The first hitting time and place of a half-line by a biharmonic pseudo process, Japan. J. Math., 23 (1997), 235–280; Math. Review 99f:60080. [15] Schwartz, L., Mathematics for the physical sciences, 2-nd ed., Hermann, 1966; Math. Review 34 #7309. [16] Strichartz, R., A Guide to Distribution Theory and Fourier Transforms, CRC Press, 1994; Math. Review 95f:42001. [17] Temam, R., Infinite-Dimensional Dynamical Systems in Mechanics and Physics, 2-nd ed., Appl. Math. Sci. 68, Springer–Verlag, 1997; Math. Review 98b:58056. 27